Spark-X2.6

Welcome to the GitHub repository of Spark-X2.6 open model series. You can find official information about Spark-X2.6, and post your questions here(Issues).

Frequently Asked Questions

See the Spark-X2.6 FAQ / 常见问题 for image-input support, local runtime choices, DSpark requirements and final-result validation.

Introduction

Building on the Spark-X2.5, we introduce Spark-X2.6-4B&1.7B, the latest generation of our compact, general-purpose language models for on-device deployment. Spark-X2.6 retains exactly the same architecture as Spark-X2.5, achieving its capability gains through improved post-training. It delivers substantial improvements in coding, agent, long-context, multilingual, reasoning, etc. These advances enable more capable and practical on-device AI for developers and users tackling real-world tasks.

Technical Highlights:

• Stronger Coding, Agentic, and Long-Context Capabilities: The models deliver substantial improvements in agentic workflows involving complex tool use, multi-step planning, and software engineering. They achieve state-of-the-art performance among similarly sized open-source models, matching or even surpassing models with approximately 9B parameters.
• Controllable Thinking Budget: The models provide three configurable thinking levels—none, low, high—allowing applications to balance reasoning quality, latency, and inference cost dynamically.

Release

• [2026/10/27] 🚀 Global launch of the Spark-X2.6-1.7B.
• [2026/10/21] 🚀 Global launch of the Spark-X2.6-4B on Hugging Face, ModelScope, Ollama.
• [2026/09/06] 🤝 Added native model architecture support for Spark‑X2.5 & X2.6 (Spark2_5ForCausalLM) in llama.cpp.
• [2026/09/01] 🚀 Global launch of the Spark-X2.5 model series on Hugging Face, ModelScope, Ollama, Modelers, and SCNet.

Model Downloads

Spark-X2.6-4B&1.7B is available on the following platforms. Choose the most suitable download source for your region and environment:

Platform	Download	Description
Hugging Face	Spark-X2.6 in Huggingface	Official Hugging Face model collection for Spark-X2.6
ModelScope	Spark-X2.6 in ModelScope	Recommended download source for users in China
Ollama	Spark-X2.6 in Ollama	Download and run the model locally with Ollama

Benchmarks

We evaluate our models and compare them with leading on-device models of similar size across a broad range of tasks, including agent, code, math, general and knowledge.

Spark-X2.6-4B

Spark-X2.6-4B achieves best-in-class performance among models of comparable size, while matching or exceeding the performance of larger 9B models on several benchmarks.

Spark-X2.6-1.7B

Spark-X2.6-1.7B surpasses models of comparable size on most benchmarks, delivering state-of-the-art results.

Edge Inference

Spark-X2.6-4B

Spark-X2.6-4B delivers inference efficiency on par with or better than comparable on-device models, including Qwen3.5-4B and Gemma-E4B. It combines this efficiency with stronger capabilities, making it a practical choice for responsive, resource-conscious on-device AI.

Spark-X2.6-1.7B

Spark-X2.6-1.7B delivers leading decoding throughput at shorter context lengths among the on-device models we evaluated, including Qwen3.5-2B, MiniCPM-2B, and LFM2.5-2.6B. Moreover, our optimized llama.cpp implementation(https://github.com/XHToken/llama.cpp) amplifies this performance advantage across all tested context lengths on the RTX 4090, underscoring its strong efficiency for responsive local AI.

Quickstart

Transformers

Install the required version with pip install transformers==4.57.1.

from transformers import AutoModelForCausalLM, AutoTokenizer
model_id = "XHToken/Spark-X2.6-4B"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    trust_remote_code=True,
    torch_dtype="auto",
    device_map="auto",
)
messages = [{"role": "user", "content": "What is the capital of Anhui Province?"}]
inputs = tokenizer.apply_chat_template(
    messages,
    tokenize=True,
    add_generation_prompt=True,
    enable_thinking=True,
    return_dict=True,
    return_tensors="pt",
).to(model.device)
outputs = model.generate(
    **inputs,
    max_new_tokens=131072,
    do_sample=True,
    temperature=1.0,
    top_p=0.95,
    top_k=0,
)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True))

SGLang

The examples below serve a local Spark-X2.6-4B checkpoint. Set MODEL_PATH to its absolute path before starting a container:

export MODEL_PATH=/absolute/path/to/Spark-X2.6-4B

Install SGLang

Use the pre-built image that tracks the Spark-X2.6 runtime:

For NVIDIA GPUs:

docker pull lmsysorg/sglang:nightly-dev-cu13-20260827-20621aa1

For Ascend NPUs:

# A3 daily build
export SGLANG_IMAGE=quay.io/ascend/sglang:main-cann9.0.0-a3

# A2 daily build (use this instead on A2 hardware)
export SGLANG_IMAGE=quay.io/ascend/sglang:main-cann9.0.0-910b

docker pull "$SGLANG_IMAGE"

Run Inference

The following commands start an OpenAI-compatible API server configured for a maximum context length of 1,048,576 tokens. This setting requires sufficient device memory; reduce --context-length when necessary.

Server

NVIDIA GPU:

docker run --rm -it \
  --gpus '"device=0"' \
  --ipc=host \
  -p 30000:30000 \
  -v "$MODEL_PATH:/root/Spark-X2.6-4B:ro" \
  lmsysorg/sglang:nightly-dev-cu13-20260827-20621aa1 \
  python -m sglang.launch_server \
    --model-path /root/Spark-X2.6-4B \
    --served-model-name spark2.6 \
    --tool-call-parser spark25 \
    --reasoning-parser qwen3 \
    --tp-size 1 \
    --mem-fraction-static 0.8 \
    --context-length 1048576 \
    --chat-template /root/Spark-X2.6-4B/chat_template.jinja \
    --host 0.0.0.0 \
    --port 30000

Ascend NPU:

docker run -it --rm -e ASCEND_USE_FIA=1 --network=host --ipc=host --shm-size=16g \
    --device=/dev/davinci0 --device=/dev/davinci1 --device=/dev/davinci2 --device=/dev/davinci3 \
    --device=/dev/davinci4 --device=/dev/davinci5 --device=/dev/davinci6 --device=/dev/davinci7 \
    --device=/dev/davinci8 --device=/dev/davinci9 --device=/dev/davinci10 --device=/dev/davinci11 \
    --device=/dev/davinci12 --device=/dev/davinci13 --device=/dev/davinci14 --device=/dev/davinci15 \
    --device=/dev/davinci_manager \
    --device=/dev/devmm_svm \
    --device=/dev/hisi_hdc \
    --volume /usr/local/sbin:/usr/local/sbin \
    --volume /usr/local/Ascend/driver:/usr/local/Ascend/driver \
    --volume /usr/local/Ascend/firmware:/usr/local/Ascend/firmware \
    --volume /etc/ascend_install.info:/etc/ascend_install.info \
    --volume /var/queue_schedule:/var/queue_schedule \
    --volume ~/.cache/:/root/.cache/ \
    --volume "$MODEL_PATH:/root/Spark-X2.6-4B:ro" \
    --entrypoint=python \
    "$SGLANG_IMAGE" \
    -m sglang.launch_server \
      --model-path /root/Spark-X2.6-4B \
      --served-model-name spark2.6 \
      --tool-call-parser spark25 \
      --reasoning-parser qwen3 \
      --tp-size 1 \
      --mem-fraction-static 0.8 \
      --context-length 1048576 \
      --chat-template /root/Spark-X2.6-4B/chat_template.jinja \
      --host 0.0.0.0 \
      --port 30000

Client

Thinking is enabled by default by both the chat template and the Qwen3 reasoning parser. To disable thinking for a specific request, set "chat_template_kwargs": {"enable_thinking": false}.

curl -s http://localhost:30000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "spark2.6",
    "messages": [
      {
        "role": "user",
        "content": "安徽的省会在哪里？"
      }
    ],
    "max_tokens": 131072,
    "temperature": 1,
    "top_k": -1,
    "top_p": 0.95,
    "repetition_penalty": 1,
    "presence_penalty": 0,
    "frequency_penalty": 0
  }'

vLLM

Deploy vLLM

vLLM provides an official Docker image for NVIDIA GPU deployment:

docker run --rm --gpus all \
  --ipc=host \
  -p 30000:30000 \
  -v "$MODEL_PATH:/models/Spark-X2.6-4B:ro" \
  vllm/vllm-openai:latest \
  --model /models/Spark-X2.6-4B \
  --port 30000 \
  --trust-remote-code \
  --served-model-name spark25 \
  --tensor-parallel-size 1 \
  --gpu-memory-utilization 0.7 \
  --enable-prefix-caching \
  --chat-template /models/Spark-X2.6-4B/chat_template.jinja

For Ascend NPUs, choose an official image for the fastest setup.

Ascend A2:

export IMAGE=quay.io/ascend/vllm-ascend:nightly-main
docker pull "$IMAGE"

export DEVICE=/dev/davinci0
export MODEL_CACHE="${HOME}/.cache"

mkdir -p "$MODEL_CACHE"

docker run --rm \
    --name vllm-ascend \
    --shm-size=1g \
    --device "$DEVICE" \
    --device /dev/davinci_manager \
    --device /dev/devmm_svm \
    --device /dev/hisi_hdc \
    -v /usr/local/dcmi:/usr/local/dcmi \
    -v /usr/local/bin/npu-smi:/usr/local/bin/npu-smi \
    -v /usr/local/Ascend/driver/lib64/:/usr/local/Ascend/driver/lib64/ \
    -v /usr/local/Ascend/driver/version.info:/usr/local/Ascend/driver/version.info \
    -v /etc/ascend_install.info:/etc/ascend_install.info \
    -v "$MODEL_CACHE:/root/.cache" \
    -p 8000:8000 \
    -it "$IMAGE" bash

Ascend A3:

export IMAGE=quay.io/ascend/vllm-ascend:nightly-main-a3
docker pull "$IMAGE"

export DEVICE0=/dev/davinci0
export DEVICE1=/dev/davinci1
export MODEL_CACHE="${HOME}/.cache"

mkdir -p "$MODEL_CACHE"

docker run --rm \
    --name vllm-ascend \
    --shm-size=1g \
    --device "$DEVICE0" \
    --device "$DEVICE1" \
    --device /dev/davinci_manager \
    --device /dev/devmm_svm \
    --device /dev/hisi_hdc \
    -v /usr/local/dcmi:/usr/local/dcmi \
    -v /usr/local/bin/npu-smi:/usr/local/bin/npu-smi \
    -v /usr/local/Ascend/driver/lib64/:/usr/local/Ascend/driver/lib64/ \
    -v /usr/local/Ascend/driver/version.info:/usr/local/Ascend/driver/version.info \
    -v /etc/ascend_install.info:/etc/ascend_install.info \
    -v "$MODEL_CACHE:/root/.cache" \
    -p 8000:8000 \
    -it "$IMAGE" bash

Ascend 950DT:

export IMAGE=quay.io/ascend/vllm-ascend:nightly-main-a5
docker pull "$IMAGE"

export MODEL_CACHE="${HOME}/.cache"

mkdir -p "$MODEL_CACHE"

docker run --rm \
    --name vllm-ascend \
    --net=host \
    --shm-size=1g \
    --device /dev/davinci0 \
    --device /dev/davinci_manager \
    --device /dev/devmm_svm \
    --device /dev/hisi_hdc \
    -v /usr/local/dcmi:/usr/local/dcmi \
    -v /usr/local/Ascend/driver/tools/hccn_tool:/usr/local/Ascend/driver/tools/hccn_tool \
    -v /usr/local/bin/npu-smi:/usr/local/bin/npu-smi \
    -v /usr/local/Ascend/driver/lib64/:/usr/local/Ascend/driver/lib64/ \
    -v /usr/local/Ascend/driver/version.info:/usr/local/Ascend/driver/version.info \
    -v /etc/ascend_install.info:/etc/ascend_install.info \
    -v "$MODEL_CACHE:/root/.cache" \
    -it "$IMAGE" bash

Install the Spark plugin inside the container:

pip install uv
uv venv ~/spark2_6
source ~/spark2_6/bin/activate
git clone https://github.com/XHToken/Spark-plugin.git
cd ./Spark-plugin
uv pip install .

Server

vllm serve "/models/Spark-X2.6-4B" \
 --port "30000" \
 --trust-remote-code \
 --served-model-name spark25 \
 --tensor-parallel-size 1 \
 --gpu-memory-utilization 0.7 \
 --enable-prefix-caching \
 --chat-template /models/Spark-X2.6-4B/chat_template.jinja

Client

curl -s http://127.0.0.1:30000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "spark25",
    "messages": [{"role": "user", "content": "What is the capital of Anhui Province?"}],
    "temperature": 1.0, 
    "top_k": -1,
    "top_p": 0.95
  }'

MLX

Spark-MLX-LLM runs the original Spark-X2.6 Hugging Face checkpoints locally. It supports Apple silicon GPU, Linux CPU, and NVIDIA CUDA on Linux. No GGUF conversion is required.

Installation

git clone https://github.com/XHToken/Spark-MLX-LLM.git
cd Spark-MLX-LLM

python3 -m venv .venv
source .venv/bin/activate

# Apple silicon
python -m pip install -e .
# Linux CPU
python -m pip install -e '.[cpu]'
# Linux with CUDA 12
python -m pip install -e '.[cuda12]' 
# Linux with CUDA 13
python -m pip install -e '.[cuda13]'

Run Inference

spark-mlx-generate \
  --device gpu \
  --dtype bfloat16 \
  --model XHToken/Spark-X2.6-4B \
  --prompt "What is the capital of Anhui Province?" \
  --max-tokens 512 \
  --temp 0

llama.cpp

[!NOTE] Native spark2_5 support requires llama.cpp b10828 or later.

Run the model in the CLI:

llama cli -m /path/to/Spark-X2.6-4B-GGUF

Launch the OpenAI-compatible API server:

llama serve -m /path/to/Spark-X2.6-4B-GGUF

Ollama

1. Download and install Ollama from ollama.com (v0.34.2 or later).

2. Run the model:

ollama run SparkLLM/Spark-X2.6-4B

LM Studio

1. Download and install LM Studio from lmstudio.ai (0.4.0 or later).

2. Search for "Spark-X2.6" in LM Studio and download the model.

3. Load the model and start chatting.

Run with the lms cli

# List available models
lms ls

# Run chat with Spark-X2.6
lms chat spark-x2.6

Unsloth

Download Unsloth Studio

[!NOTE] The latest Unsloth Studio supports native Spark-X2.6 GGUF inference.

Fine-Tuning

We recommend using Llama-Factory to fine-tune the model.

License

The Spark-X2.6 model series is licensed under the Apache 2.0 License.

Citation

If you find our work helpful, feel free to give us a cite.

@misc{sparkx2.6,
    title  = {Spark-X2.6-4B&1.7B: Advancing the Frontier of On-Device Agents for Real-World Tasks},
    author = {SparkLLM Team},
    year   = {2026}
}